Goedel-Architect introduces blueprint generation and iterative refinement for Lean 4 theorem proving, reaching 99.2% on MiniF2F-test and 75.6% on PutnamBench with DeepSeek-V4-Flash.
Primitive sets and von Mangoldt chains: Erd\H{o}s Problem #1196 and beyond
4 Pith papers cite this work. Polarity classification is still indexing.
4
Pith papers citing it
abstract
A set of integers is primitive if no number in the set divides another. We introduce a new method for bounding Erd\H{o}s sums of primitive sets, suggested from output of GPT-5.4 Pro, based on Markov chains with von Mangoldt weights. The method leads to a host of applications, yet seems to have been overlooked by the prior literature since Erd\H{o}s's seminal 1935 paper. As applications, we prove two 1966 conjectures of Erd\H{o}s-S\'ark\"ozy-Szemer\'edi, on primitive sets of large numbers (#1196) and on divisibility chains (#1217). The method also provides a short proof of the Erd\H{o}s Primitive Set Conjecture (#164), as well as the related claim that 2 is an ''Erd\H{o}s-strong'' prime. Moreover, the method resolves a revised form of the Banks-Martin conjecture, which has long been viewed as a unifying `master theorem' for the area.
years
2026 4representative citing papers
LeanMarathon uses four contract-scoped agents on an evolving blueprint coordinated by a two-stage orchestrator to formalize seven theorems from Erdős problems in Lean, proving 258 lemmas with no sorry across three runs.
Generalized Dirichlet eta functions are expressed as Gamma-process expectations, yielding a Riccati equation whose negative forcing proves strict concavity, log-concavity, positivity and monotonicity of the log-derivative, plus precise large-t asymptotics.
Introduces the VET framework to categorize and critique polarized AI narratives including hype, doom, denial, and normalcy.
citing papers explorer
-
Goedel-Architect: Streamlining Formal Theorem Proving with Blueprint Generation and Refinement
Goedel-Architect introduces blueprint generation and iterative refinement for Lean 4 theorem proving, reaching 99.2% on MiniF2F-test and 75.6% on PutnamBench with DeepSeek-V4-Flash.
-
LeanMarathon: Toward Reliable AI Co-Mathematicians through Long-Horizon Lean Autoformalization
LeanMarathon uses four contract-scoped agents on an evolving blueprint coordinated by a two-stage orchestrator to formalize seven theorems from Erdős problems in Lean, proving 258 lemmas with no sorry across three runs.
-
Riccati--Gamma Dynamics for Concavity and Asymptotics of Generalized Dirichlet Eta Functions
Generalized Dirichlet eta functions are expressed as Gamma-process expectations, yielding a Riccati equation whose negative forcing proves strict concavity, log-concavity, positivity and monotonicity of the log-derivative, plus precise large-t asymptotics.
-
VET: A Framework for Analyzing AI Discourse
Introduces the VET framework to categorize and critique polarized AI narratives including hype, doom, denial, and normalcy.